In this paper the detection and typification of grid structures in building groups is described. Typification is a generalization operation that replaces a large number of similar objects by a smaller number of objects, while preserving the global structure of the object distribution. The typification approach is based on three processes. First the grid structures are detected based on the so-called relative neighborhood graph. Second the detected grid structures are regularized by a least square adjustment of an affine or Helmert transformation. The third process is the reduction or simplification of the grid structure, which can be done using the same affine or Helmert transformation approach.